Add alma-oclc-lookup script

Author: dmolesUC

CHANGES.md

@@ -1,3 +1,7 @@
+# (next)
+
+- Add `alma-oclc-lookup` script.
+
 # 0.0.7.1 (24 May 2022)
 
 - Set minimum Nokogiri version to address

README.md

@@ -263,3 +263,26 @@ b1234
 9912348245906531	9912348245906531	
 b110824349	b110824349	991038544199706532
 ```
+
+
+### `alma-oclc-lookup`: 035$a OCLC number lookup
+
+The `alma-oclc-lookup` script takes one ore more record IDs (either Millennium bib
+numbers or Alma MMS IDs) and attempts to read the corresponding MARC records via
+[SRU](https://developers.exlibrisgroup.com/alma/integrations/sru/) and extract
+the OCLC number(s) from the 035$a field.
+
+#### Output format
+
+The output is tab-separated, in the form
+
+```none
+<record ID>	<OCLC number(s)…>
+```
+
+E.g.
+
+```none
+$ echo '991051353589706532' | bin/alma-oclc-lookup                                                                                        ✹ ✚main ‹ruby-2.7.5›
+991051353589706532      1097551039
+```

bin/alma-oclc-lookup

@@ -0,0 +1,78 @@
+#!/usr/bin/env ruby
+
+# ------------------------------------------------------------
+# Global setup
+
+# Don't buffer stdout or stderr
+$stdout.sync = true
+$stderr.sync = true
+
+# ------------------------------------------------------------
+# Dependencies
+
+# Require gems
+require 'bundler/setup'
+
+# Require lib directory
+unless $LOAD_PATH.include?((lib_path = File.expand_path('../lib', __dir__)))
+  puts "Adding #{lib_path} to $LOAD_PATH"
+  $LOAD_PATH.unshift(lib_path)
+end
+
+require 'berkeley_library/alma'
+require 'marc/spec'
+
+# ------------------------------------------------------------
+# Configuration
+
+# Configure Alma URLs etc.
+BerkeleyLibrary::Alma::Config.default!
+
+# Configure logging
+logger = BerkeleyLibrary::Logging::Loggers.new_readable_logger($stderr)
+logger.level = Logger::Severity::WARN
+BerkeleyLibrary::Logging.logger = logger
+
+# ------------------------------------------------------------
+# Constants
+
+OCLC_NUM_RE = /(?<=\(OCoLC\))(?<num>\d+)/.freeze
+
+# ------------------------------------------------------------
+# Utility methods
+
+# Read raw IDs from  STDIN
+def raw_ids
+  @raw_ids ||= Enumerator.new do |y|
+    $stdin.each_line(chomp: true) do |ln|
+      y << ln.strip
+    end
+  end
+end
+
+def ids_035a_for(record_id)
+  return unless record_id
+  return unless (marc_record = record_id.get_marc_record)
+
+  MARC::Spec.find('035$a', marc_record)
+end
+
+def oclc_nums_for(record_id)
+  return [] unless (ids_035a = ids_035a_for(record_id))
+
+  ids_035a.filter_map do |sf_035a|
+    next unless (val = sf_035a.value)
+    next unless (md = OCLC_NUM_RE.match(val))
+
+    md[:num]
+  end
+end
+
+# ------------------------------------------------------------
+# Main program
+
+raw_ids.each do |raw_id|
+  record_id = BerkeleyLibrary::Alma::RecordId.parse(raw_id)
+  oclc_nums = oclc_nums_for(record_id)
+  puts [raw_id, *oclc_nums].join("\t")
+end